Feature bagging for outlier detection

By Aleksandar Lazarevic, Vipin Kumar. Knowledge Discovery and Data Mining, 2005

737 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/e3116ebb52c152deae918a04c34441ac0d956b8a

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Statistics and Probability

Citations in the corpus

Outlier Detection (2008, background): “This is closely related to the standard supervised learning problem except that all the training samples have the same normal label.”
Classification and Novel Class Detection in Concept-Drifting Data Streams under Time Constraints (2011, background): “Finally, Section 6 concludes with directions to future works.”
MAD-GAN: Multivariate Anomaly Detection for Time Series Data with Generative Adversarial Networks (2019, methodology): “For example, the Angle-Based Outlier Detection (ABOD) method [19] and Feature Bagging (FB) method [20] deal with data by taking variable correlations into consideration.”
Outlier detection by active learning (2006, methodology): “(On some portion of the data sets used in [10], all of the outlier detection methods do no better than random guessing.”
A Unifying Review of Deep and Shallow Anomaly Detection (2020, background): “Ensembles have a conceptual justification in the context of AD [334], and they have been shown to work well empirically [469], [470].”
Outlier Analysis (2013, methodology): “The approach has been shown to show consistent performance improvement over full dimensional methods for many real data sets in [289].”
Learning Graph Structures With Transformer for Multivariate Time-Series Anomaly Detection in IoT (2021, methodology): “Generally, the existing approaches can roughly fall into two lines: reconstruction-based models (R-model) [10], [11], [12], [6], [13] and forecasting-based models (F-model) [14], [15], [8], [16], [17], [4].”
Incremental Local Outlier Detection for Data Streams (2007, methodology): “However, for typically used values of k (5-20) [4,5] the number ofLOF updates becomes constant for N>5000.”
A Deep Neural Network for Unsupervised Anomaly Detection and Diagnosis in Multivariate Time Series Data (2018, methodology): “Other traditional methods include correlation methods (Kriegel et al. 2012), ensemble methods (Lazarevic and Kumar 2005), etc.”
Isolation-Based Anomaly Detection (2012, methodology): “Feature Bagging [Lazarevic and Kumar 2005] is proposed to improve LOF, in which data from single attributes are used to generate multiple models and the .nal anomaly scores are aggregated from multiple models.”
Artificial intelligence based anomaly detection of energy consumption in buildings: A review, current trends and new perspectives (2021, methodology): “In [129], after deriving various feature sub-ensembles randomly from the initial feature, anomalies are identified and the performance is estimated in each sub-ensemble before fusing them to come out with the final output.”
Adversarial attacks against intrusion detection systems: Taxonomy, solutions and open issues (2013, methodology): “This approach is also known as feature bagging [90] or random subspace method [160].”
Interpreting and Unifying Outlier Scores (2011, methodology): “For comparison, we implemented the methods of [17,34,38] (see Section 4).”
Outlier Detection with Autoencoder Ensembles (2017, methodology): “Several ensemble methods such as feature bagging [12], subspace histograms [19], high-contrast subspaces [9], and locally relevant subspaces [14, 15] have been proposed.”
Outlier Detection with Kernel Density Functions (2007, background): “Recently, LOF (Local Outlier Factor) [9] and LOCI (Local Correlation Integral) [35] algorithms have been successfully applied in many domains for outlier detection in a batch mode [4, 5, 35].”
Outlier Detection for Text Data (2017, methodology): “The key methods, which are used frequently for outlier analysis include distance-based methods [13, 20], density-based methods [4], and subspace methods [2, 11, 16, 19, 15].”
ECOD: Unsupervised Outlier Detection Using Empirical Cumulative Distribution Functions (2022, methodology): “A natural way to characterize outliers is to take them to correspond to rare events that occur in low density parts of probability distribution ([25], [26]).”